PolSAR Terrain Classification Based on Transformer-Unet Hybrid Model in the Complex Domain

The Transformer model has recently achieved success in image classification tasks. Additionally, the polarimetric synthetic aperture radar (PolSAR) terrain classification has been confirmed to achieve better classification results in the complex domain compared to the real domain. Therefore, this paper introduces the Transformer into the complex domain and proposes a new classification network which named complex-valued Transformer-Unet hybrid model (CT-Unet), for PolSAR terrain classification. Its hybrid approach is to embed Transformer module into the U-Net encoder section for feature extraction on PolSAR data. The experimental results on PolSAR terrain classification using the Xi’an dataset demonstrate that our proposed model effectively improves the accuracy of PolSAR terrain classification.